About ClipCat

ClipCat started as one person's tool for turning long, messy takes into tight videos without spending an evening in a timeline.

The idea

Editing a talking-head video is mostly the same chores every time: cut the retakes and pauses, punch in on the good lines, add the screenshots, the captions and a little sound, then cut shorts and write the description. ClipCat does that first pass with the AI you already use, lays every change out on the timeline with a reason, and leaves every call to you.

How it's built

ClipCat is an Electron app for macOS and Windows. Video is rendered with ffmpeg, speech is transcribed with NVIDIA's Parakeet (through FluidAudio) or whisper.cpp, faces are found with MediaPipe, and the subject cutout uses ISNet on ONNX Runtime. The name and the cat come from the cutting room: the C is a clip with a wedge trimmed out of it.

Open source

The source is on GitHub. Parts of the app shell are adapted from Diffusion Studio's editor under the MPL-2.0, and the bundled tools keep their own licenses.

Last updated .